Linguistic or Behavioural

Question: In the case of Mary, which evidence is more compelling – linguistic or behavioural?

In the case of Mary, I am convinced that behavioral evidence is more compelling, especially if it contradicts the linguistic evidence. We mentioned deception in class and, though I am not necessarily convinced that consciousness started as a means to deceive, it's true that humans have such an ability. It's also true that the behavioural response, as it exists on a more primitive level, is more difficult to fully control to sync up with linguistic deception. If a person says that they are not in pain, yet they scream, contort their faces, and grit their teeth, it is probably safe to assume that they are indeed in some sort of distress. This, I suppose would not be the case, if they proceeded to tell you that they had a number of disorders that caused them to do all of those things even when they are not in pain. 

It seems to me that descriptive and explanatory linguistic accounts may be more compelling, in many cases, than behavioural actions. Still, a simple linguistic account may not suffice to explain away behavioural actions. I do not know where one would have to draw the line; I suppose this would highly depend on each individual case.